Am Montag, 17. April 2017 um 20:31:07, schrieb Tommaso Cucinotta 
<[email protected]>
> On 17/04/2017 12:50, Kornel Benko wrote:
> > Am Montag, 17. April 2017 um 12:39:26, schrieb Kornel Benko <[email protected]>
> >>> Perhaps the issue is just that we terminate LyX with "kill -9" in 
> >>> keytest.py [1]. Would you please try replacing that "-9" with "-SIGQUIT" 
> >>> or similar, and see what happens ? (after rolling back the \Ax, which, 
> >>> from what I'm suspecting, are just adding some extra time before the kill 
> >>> -- probably I had this issue as well, because that "sleep(0.5)" I'm 
> >>> seeing below is also equally suspicious :-), but who remembers...
> >>
> >> Now lyx never ends
> >> I have to do it manually (File->Esci)
> >
> > And the test fails because the cursor position is now saved and used in 
> > next session.
> > (Previously it was not saved becase of kill -9)
> 
> so, tests reusing the same .lyx file should all begin with a \C[Home], as you 
> never know when such cur pos might be saved ...
> 
> Btw, instead of ending each test with \Ax, what about ending them with \Cq ? 
> (shortcut for File->Quit) ?

Or \Axlyx-quit\[Return], this is language independent. OTOH, maybe we have to 
wait before sending it.
I tried, and lyx often missed some keystrokes ...

The problem here is that the buffer may be dirty. E.g. you get a dialog for 
save, discard, or abort.

And what if lyx is in an endless loop? (Should not be, but who knows)

> But, the current keytest.py expects LyX to be found running on TestEnd, or it 
> fails, so an explicit \Cq in the -in.txt file would not work.
> Instead, what about replacing the kill -9 with an attempt of clean exit ? 
> e.g., see the attachment.

Nice, but we should wait some seconds before using kill -9 the first time

> Thanks,
> 
>       T.

Your patch works good for this test.

Attachment: signature.asc
Description: This is a digitally signed message part.

Reply via email to